A deployment model describes how computing resources are provided and shared; a service model describes which capabilities a provider operates. Physical placement adds a separate question: where the resources run.

Deployment and service responsibilities

Public cloud and private cloud describe different deployment arrangements. Under the NIST definition, private cloud infrastructure serves the exclusive use of one organization. Owning a room of servers does not by itself establish the cloud characteristics in that definition.

Software as a service, or SaaS, supplies an application operated by a provider. Infrastructure as a service, or IaaS, supplies infrastructure resources on which the customer can operate applications. These labels describe different divisions of work. Neither label alone identifies every physical site or subcontractor involved.

An application vendor and an infrastructure provider can be different organizations. An application can also depend on customer-operated identity, network and data systems. Describing only the supplier that sends the invoice leaves those dependencies unresolved.

A bank’s risk code and collateral application

Take a bank running its own risk code on rented compute and buying a collateral application as a service. The bank operates the first application. The supplier operates the second application within the agreed service boundary. Both arrangements involve external infrastructure or software, but the responsibilities differ.

The bank’s own code still needs controlled releases, numerical configuration and operational monitoring. The SaaS application still needs the bank’s agreement data, eligible collateral inventory, permissions and reconciliation with its other records. Provider operation does not determine which financial action the bank has authorized.

Deutsche Bank’s documented combination of SAP finance workloads on Google Cloud and hybrid support for its Autobahn FX platform illustrates coexistence at the workload level. A description of one migrated workload does not assign the same placement to every system at the institution.

Mixed estates and migration boundaries

A mixed technology estate can include dedicated trading servers, mainframes, cloud resources and SaaS applications. NIST’s formal hybrid-cloud definition concerns connected distinct cloud infrastructures; a mixed estate can contain additional systems outside that definition.

Migration changes a component’s placement or operator. Its financial interfaces still need to preserve identifiers, transaction states and recovery behavior. Moving a ledger’s storage does not automatically move the authority to correct a posting.

The useful description of a financial deployment names the workload, physical placement, resource-sharing model, application operator and retained customer responsibilities. What changes between arrangements is the division of operating work and the dependency structure.

Questions about IaaS

Is private cloud the same as owning servers?

No. Private cloud describes an exclusively used cloud environment; server ownership alone does not define the service characteristics.

Does SaaS specify which cloud provider runs the application?

No. SaaS describes the application service boundary; infrastructure placement needs a separate description.